By the end of the mission, and thus also at the end of the
mission’s funds, the individual image strips, each up to
220,000 rows of pixels at 300 pixels per row, with each
pixel representing an area of 75 m x 75 m on the surface,
had been assembled into mosaics. Table 1 summarizes the
SAR image products obtained from the mission, Table 2
addresses the range of data products in addition to those
derived from the images. It becomes evident that the entire
globe is covered by image mosaics at a scale up to 27
times smaller than that at which raw images exist. The full
resolution products, however, were only assembled from
perhaps 20% of the images, and over 10% of the surface.
F-MIDRs
Full-Resolution Mosaicked Image 5°x 5° 75m pixels
Data Records
C1-MIDRs
Averaging 3x3 pixels to a reduced resolution
Mosaicked Image Data Record 15° x 15° 225m pixels
C2-MIDRs
Averaging 9 x 9 pixels 45° x 45° 625m pixels
C3-MIDRs
Averaging 27 x 27 pixels 120° x 120° 2025m pixels
F-MAPs
Reprocessed Full-Resolution Mosaics 75m pixels
Table 1: Summary of the major data products derived from
Magellan SAR images. The C1-, C2-, C3-MIDR sets all cover the
entire planet, devided into mapsheets of given extent
Radar Altimetry
ARCCDR Altimetry and Radiometry Data Records Individual echoes
G-TDR Global Topography Data Records 5 km grid
G-SDR Global Slope Data Records 5 km grid
Radiometry
G-REDR Global Reflectivity Data Records 5 km grid
G-EDR Global Emissivity Data Records 5 km grid
Gravity
High Resolution Gravity Map
Table 2: Summary of other data products obtained from the
Magellan mission in addition to the images.
Magellan challenged the data processing capabilities
available during the mission. Still, all the 400 Gbytes of
raw images are now available in NASA's planetary data
system PDS, being responsible for data from all US
planetary missions. After being transferred to an up to date
storage media, also the raw radar phase and amplitude
signals are planned to be managed by PDS.
Even a few years after completion of the mission itself, the
data set is still large enough by contemporary standards to
stimulate one's imagination; in addition, the uniform
coverage of an entire planet without the concern for
national borders helps to provide Magellan's images a
function of „role model" for managing terrestrial remote
sensing data sets.
Many of the procedures used during the mission to create
image products and information about the planet's surface
were, in essence, ,quick", not necessarily most accurate
and optimum. Since the mission's pressures and priorities
have now faded one can take the time to develop a strategy
for producing optimum data products and most accurate
descriptions of the surface topography.
We report on ongoing work and will show how laid-over
terrain features and dissimilarities between overlapping
images are important elements in a strategy to reprocess
the raw images. Of course, the sheer size of the data set
must itself be a focus of recent work to prepare for an
efficient analysis of Magellan imagery.
492
2 THE PLANETARY DATA SYSTEM
2.1 Organizational Aspects
Extraterrestrial space missions result in data that are being
maintained by a NASA-funded organization of 8 research
centers, half academic, half governmental. Each represents
a functional node of a so-called network denoted as
Planetary Data System (PDS). Its headquarter is at the Jet
Propulsion Laboratory in California. PDS deals with
images, atmospheric, gravity and other data resulting from
past missions to other planets and moons, not, however,
with data of the Earth. The Magellan images themselves
are the responsibility of the so-called ,,Geosciences Node"
at Washington University (St. Louis). This node in turn
manages a European Magellan Data Node at the Technical
University in Graz. Magellan images are also being
studied at the ,Imaging Node" at the Center for
Astrogeology in Flagstaff, Arizona, which is part of the
US Geological Survey (USGS). PDS is networked via the
Internet. Access is free for anyone on the Internet via the
address http:/www.pds.nasa.gov/. PDS maintains an on-
line facility for those interested to search a data catalog.
Images can be ordered and are being distributed by mail.
2.2 Status of Magellan Data Products
There is currently only one systematic effort going on to
process so-called F-MAPs, full-resolution mosaics
produced at the USGS Center for Astrogeology, one sheet
per 12° x 12°, and created from the original radar image
strips. Those images are corrected for the topographic
relief known from altimetry. The individual image strips
are connected by tie points, but the result is not corrected
for detailed topography, even though such information
would be available if a stereo effort were undertaken. The
entire planet is expected to be covered by 1998. No
funding exists at this time to process all 3 imaging cycles
of Magellan. Instead a complete coverage of the entire
planet is the goal so that each location of the planet is at
least available on one F-MAP. This database is meant to
replace previous F-MIDRs (see Table 1), the full-
resolution mosaicked data records created during the
Mission.
Data products suffer from the limitations that exist during
a mission, namely insufficiently known topography and
the priority of speed over accuracy and quality.
Weaknesses in those data products exist:
e The ephemeris on which the image products are based
is the predicted one. An ephemeris obtained by post-
mission processing of all observations and by
consideration of superior gravity data does not exist.
To improve the data products, one would have to first
reprocess the ephemerides, then reprocess the raw
radar echo signals with the improved satellite's
ephemeris.
e The image products are not corrected for topographic
relief as it resulted from the mission, nor are they
corrected for topographic relief as it could be extracted
from the images themselves where stereo-overlaps
exist (see Leberl, 1993a,b) The individual data
products are not terrain corrected and will thus not fit
one-another when covering the terrain from opposite
sides.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996
*-9 "T PS — PM "S. —4— t3 —beà £o ^ A 3
